Generate 3D human motion videos from text prompts
Retrieve 3D human motion videos from text descriptions